Recommendations — what works, at what cost, for which voters

How to read this page

The findings on the site each name part of the problem. This page lines up the fixes.

• Finding 4 (why registered young people didn't vote) and Finding 3 (why others didn't register) name the barrier.
• Finding 5 (how young voters actually vote) shows which delivery method works when their state offers it.
• Finding 6 (state policy as the lever) shows which states have built that delivery method and which haven't.
• The size of the lift when you expand a method is what foundations and election administrators are buying. The cost per new voter is the unit price.

Across 13 federal elections (2000-2024), about half of registered young people aged 18-29 who didn't vote say it was time and logistics. About a quarter say engagement. About 9% say access. So the biggest returns come from policy reform and administrative changes that help voters finish the ballot — not from generic get-out-the-vote campaigns.

Group-level evidence caveat

The intervention catalog records effect ranges at the population level, with a documented youth multiplier where the primary source publishes one. It does not contain race × gender × age group-specific effect sizes for most interventions. Where a youth multiplier exists, we use it; where it does not, we report the general-population range and flag "general youth 18-29 — group-level effect not separately published." This is a genuine limitation of the evidence base, not a gap in this analysis. Finding 5 pairs this catalog with Current Population Survey (CPS) group-level method preference data to narrow the applicable group; the catalog alone cannot.

Recommendations matrix

Policy reform (state-level; multi-cycle compounding)

Caveat: these are statute-level changes; per-vote cost amortizes once passed. Initial passage campaign costs are borne separately and not reflected above. Pre-registration carries a multi-cycle lag; it does not deliver near-term gap closure.

Administrative improvement (election-office-level)

Caveat: effects are marginal where baseline administration is already strong. Value is often in confidence and equity-of-access, not net turnout.

Programmatic GOTV (campaign-cycle; per-contact costs)

Caveat: effects depend on genuine peer relationship (SMS, relational); do not substitute broadcast for peer-sent. 501(c)(3) partners must scope door-to-door and phone banking as civic education, not electioneering. Field capacity is finite; scaling past local density produces diminishing returns.

Civic infrastructure (multi-cycle, long-lag)

Caveat: documented on program alumni; cannot generalize to non-participating youth. Long lag precludes attribution to any near-term cycle.

Institutional reform (structural; multi-cycle compounding)

Caveat: governance/competitiveness framing only; partisan consequences out of scope. RCV general-election effect smaller than primary-election effect.

What not to fund

If these are funded for other reasons (brand-building, organizational sustainability, fundraising), that rationale should be stated explicitly rather than claimed as voter mobilization.

Funder-fit matrix

22 funders × 6 intervention categories. C = category fit; P = partial fit; blank = out of category. NP column: Y = fully nonpartisan 501(c)(3); L = nonpartisan with positions; M = mixed structure; N = 501(c)(4) / electoral vehicle.

Seventeen of twenty-two are 501(c)(3)-compatible. State Voices Action, Way to Win, Arabella network, Alliance for Youth Action, and some modes of Common Cause are 501(c)(4) or mixed — they appear in the matrix because they fund the correct category, not because they are partnership candidates for a 501(c)(3) deliverable. Government entities (AmeriCorps, EAC, California SoS, Colorado SoS, NSF) require procurement-compatible scoping with distinct disclosure requirements.

How to read this page alongside the findings

• Finding 3 answers which access barrier is cited by which group; this page says what closes it.
• Finding 4 answers which logistical barrier is cited by which group; this page says what closes it.
• Finding 5 is the connective tissue: revealed group-level method preference determines whether an intervention applies to that group.
• Finding 6 tells you which states already have the lever in place; the intervention rows here tell you the expected effect where it is expanded.
• Finding 7 is the institutional-reform section of the same logic, with within-state pre/post identification as a planned extension.

If a row in this matrix does not trace back to a group-level finding via one of those pages, treat it as general-population evidence and flag it in the grant narrative. The trust-but-verify rule applies here too.